SEARCH final project meeting: Palaeoclimate theme outcomes Dr Joelle Gergis, Dr Ailie Gallant, Dr Raphael Neukom and Prof David Karoly Placing the 13-year ‘Big Dry’ in a long-term context ? Recent decadal-scale drying – is it unusual in a longer-term context? – Only have 100 years of instrumental data – Difficult to assess decadal-scale variability using short time series The role of palaeoclimatology Tree growth rings Biological and geological indicators capture natural climate variability on seasonal–centennial timescales e.g. tree rings, corals, ice cores, lake sediments, cave records. Science known as ‘palaeoclimatology’ SEARCH used ‘high resolution’ i.e. monthly– annually resolved palaeoclimate records to extend the instrumental climate record centuries into the past Ice laminations Provides estimates of pre-industrial or ‘natural’ climatic variations to assess recent climate extremes Palaeoclimate reconstructions are a tool for comparison with climate models: Assess the role of ‘natural forcing’ e.g. solar, volcanic, internal ocean– atmospheric processes (ENSO, SAM, IOD) and anthropogenic greenhouse gas forcing Coral banding SEA rainfall reconstruction pilot study (1783–1988) Corals Trees Ice cores • ‘Proof of concept’ using 12 well-dated, annually resolved records from Australasia with published climate sensitivity (largely ENSO) to assess the feasibility of developing a rainfall reconstruction for SE Australia • Locations sensitive to large-scale climate modes: El Niño–Southern Oscillation, Indian Ocean Dipole, Southern Annular Mode Gergis et al. (2012), Climatic Change Using teleconnections to infer regional rainfall variations • Unfortunately there are no records except Tasmanian tree rings are available from the SE Australian mainland • Using remote teleconnections to estimate regional rainfall variability • Can a common climate signal extracted from teleconnected regions adequately represent variations in southeast Australian rainfall? Using remote teleconnections to estimate regional rainfall variability Can we reconstruct southeast Australian rainfall using remote observational data? • Used the Global Historical Climate Network (GHCN) of observational stations • Closest to palaeoclimate data locations • 30 years of data common to all records • Assessed responses from different variables – temperature, precipitation, pressure and seasurface temperatures Reconstructing May–April southeast Australian rainfall from remote observations r =0.69 Annual Variance explained by the remote observational network 47% r =0.92 Decadal 85% Gergis et al. (2012), Climatic Change Capturing large-scale, coherent rainfall variations: (i) instrumental and (ii) palaeoclimate data GHCN observations Palaeo records -Grid points where a statistically significant proportion of the variations in May–April Australian rainfall can be represented by 12 station GHCN network (left) and first 3 Principal Components of the palaeoclimate network (right) - Palaeo network is able to capture the common signal seen in the instrumental GHCN station network Gergis et al. (2012), Climatic Change Projecting coherent signals onto SEA rainfall Multiple linear regression to project the three leading palaeo PCs onto southeast Australian rainfall Independent ‘calibration’ and ‘verification’ periods to determine regression coefficients (αn) x1 rainfall = x2 α1x 1 + α 2x 2 + α 3x 3 x3 Gergis et al. (2012), Climatic Change Sensitivity of ‘calibration’ and ‘verification’ periods Inter-annual calibration and verification metrics Model skill sensitive to choice of calibration and verification period Statistic Definition ar2 Variance explained adjusted for degrees of freedom r Correlation RE Reduction of Error CE Coefficient of Efficiency ST Sign Test RMSE Root Mean Square Error Gergis et al. (2012), Climatic Change Histograms of skill metrics provide better estimates of skill than traditional single estimates Monte Carlo resampling to generate a 10 000-member ensemble to better quantify uncertainty: – select randomised decades for calibration over 1900–1988 – remaining years used for model verification Gergis et al. (2012), Climatic Change Ensemble median to represent our ‘best estimate’ SEA May–April rainfall reconstruction Captures 33% of inter annual variations in observed SEA rain Statistic Definition ar2 Correlation adjusted for degrees of freedom r Correlation RE Reduction of Error CE Coefficient of Efficiency ST Sign Test RMSE Root Mean Square Error rcalibration 0.85 ± 0.15 Median of 10,000 member ensemble captures 72% of decadal variations in instrumental rainfall ar2calibration rverification 0.70 ± 0.25 0.85 ± 0.19 RE CE 0.73 ± 0.32 0.66 ± 0.52 ST 80% ± 20% RMSE 0.56σ ± 0.19σ Gergis et al. (2012), Climatic Change SE Australian rainfall variability: 1783 – 2009 Wettest period in instrumental record (1950s and 1970s) Driest period in instrumental ? Rapid changes from wet torecord dry (and to wet) 2000s) (latedry 1990s/early evident through record Median of 10,000 member ensemble of rainfall reconstruction Observed area-averaged rainfall from AWAP rainfall grid (1900–2009) Observed area-averaged rainfall from nine high-quality stations (1873–2006) Gergis et al. (2012), Climatic Change How does the ‘Big Dry’ compare to rainfall estimations since 1783? 2.9% According to our rainfall reconstruction ensemble, there is a 97% chance that the 1998–2008 decadal rainfall deficit was the lowest since European settlement Gergis et al. (2012), Climatic Change Unusual happenings: 1820s and 1830s Independent records from Lake George report maxima in lake levels during 1820s – ‘A magnificent sheet of water’ (H. C. Russell, 1821) – ‘20 miles long and 8 miles wide’ (H. C. Russell, 1824) Gergis et al. (2012), Climatic Change Low frequency ENSO–SE Australian rainfall relationship • Comparing low-frequency variations in reconstructed ENSO to assess the stability of southeast Australian rainfall • McGregor et al. (2010) looked for a common signal in a number of ENSO reconstructions back to A.D. 1650 – proxy for decadal-scale variations in the Pacific (IPO) Interdecadal Pacific variability and southeast Australian rainfall: 1793–1970 Gergis et al. (2012), Climatic Change • • Breakdown is associated with pronounced wet period 1818–1833 (peaking at ~100mm above C20th average) is a real climate signal not a data issue (tested individual proxies, combinations and independent ENSO reconstructions)..decoupling of SEA rainfall–ENSO relationship (seen in other studies overseas) Pre 1840: early 1800s cooling (1oC colder than present), and period of high tropical volcanism e.g. 1816 Tambora eruption ‘year without a summer’…opportunity for pre-industrial detection and attribution studies River Murray streamflow reconstruction, 1783-1988 Gallant and Gergis (2011), Water Resources Research r=0.49 r=0.72 24% of annual variations captured 52% of decadal variations captured - Nine proxies used for August–July streamflow reconstruction (none from within the MDB) - Losing reconstruction skill at the catchment level, but the results are promising - Reconstruction can only capture rainfall component of streamflow so is limited River Murray streamflow and the IPO Gallant and Gergis (2011), Water Resources Research Recent River Murray streamflow deficits in a longerterm context • Where does the River Murray streamflow deficit sit in a longer-term context? River Murray streamflow Using RECONSTRUCTED OBSERVATIONS from the 10,000member reconstruction ensemble we estimate that there is only a 2.3% chance that the 1998–2009 streamflow deficit has been exceeded since 1783 (length of the reconstruction) Also used SIMULATED statistical modeling of River Murray streamflow (100,000 year synthetic simulations based on parameters derived from our 10,000 palaeostreamflow reconstructions) to estimate that the 1998–2009 streamflow deficit has an Average Recurrence Interval of 1 in 1500 years Gallant and Gergis (2011), Water Resources Research Estimating hydroclimatic variations in Melbourne’s water supply catchments using palaeoclimate data Ailie Gallant, Joelle Gergis and David Karoly With thanks to K.S. Tan, Bruce Rhodes and Ted Chylinski Instrumental and palaeoclimate data Melbourne Catchment – Maroondah, O’Shannassy, Upper Yarra and Thomson S11 streamflow network iS7 independent streamflow network R9 rainfall network Melbourne Catchment Streamflow Melbourne Catchment Rainfall Preliminary results: extremes in a long-term context Extreme streamflows in reconstruction are consistent with magnitude and frequency of observed streamflows during very wet and very dry years, and very wet decades. Observed streamflows during very dry decades (i.e. 1997-2007) unusual in reconstruction – 0.5th percentile. Average Return Intervals (ARI) sensitive to input parameters – difference in mean of < 5% leads to order of magnitude difference in ARI PAGES regional 2K network – Aus2K - Global effort to consolidate regional palaeo data of the last 2000 years for IPCC AR5 -To produce high-resolution, proxy-based climate reconstructions for comparison with high resolution Earth System Models -‘Aus2K’ Australasian component: important area of southern mid-latitudes: El Nino– Southern Oscillation (ENSO), Southern Annual Mode (SAM), Indian Ocean Dipole (IOD) Source: http://www.pages-igbp.org/index.html Availability of palaeoclimate data in the Australian region IPCC AR4 (2007) Only 5 annually or decadally resolve records available for the Australasian region for the last IPCC report Neukom and Gergis recently compiled and reviewed 174 high-resolution records from the Southern Hemisphere to improve data availability for climate analysis Now over 50 sites from Australasia available for palaeoclimate reconstruction Neukom and Gergis (2012), The Holocene Development of the Au2K database - Developed a database containing all monthly–annually resolved palaeoclimate records from Australasia covering last 2K (inc associated literature) - All records quality checked, reassessed (e.g. tree ring standardisation, EPS assessment etc) and reprocessed into ‘ready to use’ format for climate analysis - Searchable by domain, archive etc then data exported for analysis – research tool Reassessing the climate sensitivity of the Australasian palaeoclimate network -Reassessing Australasian proxies for potential to develop temperature and rainfall reconstructions: strong covariations in instrumental observations are found in some regions - Spatial correlation map of WA Callitris tree rings vs. JJASON (winter) rainfall and temperature - Published based on rainfall sensitivity but the record is clearly responding to winter temp in South Western Australia and Northern Territory ‘dry’ season - Identifying temperature, rainfall and ‘both’ proxy subsets for sub regional climate reconstruction Australasian temperatures of past 1000 years Gergis et al. (2011), Journal of Climate, in revision. - 1000 year Australasian spring/summer (SONDJF) mean temperature anomalies (land and ocean): mean of 3000member ensemble based on varying reconstruction parameters (28 proxy records) - 96.5% of our reconstruction ensemble members indicate that there are no other warm periods in the past millennium that match or exceed post-1950 warming observed in Australasia - Three warmest decade of the past 1000 years occur consecutively in the 1980s, 1990s and 2000s - CSIRO Mk 3L model comparison show that internal atmospheric–ocean circulation and anthropogenic forcing have more influence than solar and volcanic variations in reconstructed Australasian temperatures: greenhouse gas forcing from 1950 onward is swamping the influence of ‘natural variability’ Palaeoclimate outcomes • Palaeoclimatology is a powerful tool for assessing pre-industrial climate variations to assess recent anthropogenically forced climate change • Progress by a small team includes the development of: 1. SEA Rainfall reconstruction for SEA, A.D. 1783-2009 2. River Murray streamflow reconstruction, A.D. 1783-2009 3. Experimental Melbourne Catchment streamflow, A.D. 1783-2009 4. Southern Hemisphere palaeoclimate database for large-scale climate circulation reconstruction e.g. ENSO, SAM, IOD 5. Australasian temperature reconstruction, A.D. 1000–2000 • Recent 1998–2009 drought and low streamflow periods are anomalous in the context of the past two centuries – return period of 1 in 1500 years • Post 1950 temperatures in the Australasian region are the warmest of the past 1000 years, anthropogenic forcing is now the dominant cause of temperature variations Publications 1. Gergis, J., Gallant, J. E., Braganaza, K., Karoly, D. J., Allen, K., Cullen, L., D'Arrigo, R., Goodwin, I., Grierson, P. and McGregor, S. (2012). On the long-term context of the 1997–2009 ‘Big Dry’ in south-eastern Australia: insights from a 206-year multi-proxy rainfall reconstruction Climatic Change: 111 (3): 923–944. 1. Gallant, A. J. E. and Gergis, J. (2011). An experimental streamflow reconstruction for the River Murray, Australia, 1783–1988. Water Resources Research 47 (W00G04): doi:10.1029/2010WR009832 2. Neukom, R. and Gergis, J. (2012). Southern Hemisphere high resolution palaeoclimate records of the past 2000 years. The Holocene 5: 501–524. 3. Gergis, J., Neukom, R., Gallant, A., Phipps, S.J., Karoly, D.J. and Aus2K members. Evidence of rapid late 20th century warming from an Australasian temperature reconstruction spanning the last millennium. Journal of Climate (in revision). 1. Phipps, S., Gergis, J, McGregor, H., Gallant, A., Neukom, R., Stevenson, S., van Ommen, T., Brown, J., Fischer, M., Ackerley, D. Palaeoclimate data–model comparison: concepts and application to the climate of Australasia over the past 1500 years. Journal of Climate (in revision). 1. PAGES Regional 2K consortium including Aus2K coauthors Gergis, J., Lorrey A.M and Phipps, S.J. Continental-scale temperature variability over the Common Era, Science (in review).